Lagging Regions and Development Strategies

Transcription

1 IFPRI Discussion Paper September 2009 Lagging Regions and Development Strategies The Case of Peru James Thurlow Samuel Morley Alejandro Nin Pratt Development Strategy and Government Division and Markets, Trade, and Institutions Division

2 INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE The International Food Policy Research Institute (IFPRI) was established in IFPRI is one of 15 agricultural research centers that receive principal funding from governments, private foundations, and international and regional organizations, most of which are members of the Consultative Group on International Agricultural Research (CGIAR). FINANCIAL CONTRIBUTORS AND PARTNERS IFPRI s research, capacity strengthening, and communications work is made possible by its financial contributors and partners. IFPRI receives its principal funding from governments, private foundations, and international and regional organizations, most of which are members of the Consultative Group on International Agricultural Research (CGIAR). IFPRI gratefully acknowledges the generous unrestricted funding from Australia, Canada, China, Finland, France, Germany, India, Ireland, Italy, Japan, Netherlands, Norway, South Africa, Sweden, Switzerland, United Kingdom, United States, and World Bank. AUTHORS James T. Thurlow, International Food Policy Research Institute Research Fellow, Development Strategy and Governance Division j.thurlow@cgiar.org Samuel Morley Visiting Senior Research Fellow Markets, Trade and Institutions Division s.morley@cgiar.org Alejandro Nin Pratt, International Food Policy Research Institute Research Fellow, Development Strategy and Governance Division a.ninpratt@cgiar.org Notices 1 Effective January 2007, the Discussion Paper series within each division and the Director General s Office of IFPRI were merged into one IFPRI wide Discussion Paper series. The new series begins with number 00689, reflecting the prior publication of 688 discussion papers within the dispersed series. The earlier series are available on IFPRI s website at 2 IFPRI Discussion Papers contain preliminary material and research results. They have not been subject to formal external reviews managed by IFPRI s Publications Review Committee but have been reviewed by at least one internal and/or external reviewer. They are circulated in order to stimulate discussion and critical comment. Copyright 2008 International Food Policy Research Institute. All rights reserved. Sections of this material may be reproduced for personal and not-for-profit use without the express written permission of but with acknowledgment to IFPRI. To reproduce the material contained herein for profit or commercial use requires express written permission. To obtain permission, contact the Communications Division at ifpri-copyright@cgiar.org

3 Contents Acknowledgements v Abstract vi 1. Introduction 1 2. Regional linkages in Peru 3 3. Modeling the linkages between the coastal and inland regions 8 4. Sources of growth and divergence Policies to encourage regional convergence Conclusion 21 Appendix 22 References 27 iii

4 List of Tables Table 1. Summary statistics of coastal and inland regions 4 Table 2. Macro social accounting matrix (SAM) for Peru s coastal and inland regions (2002, millions of Soles) 5 Table 3. Macroeconomic results for growth scenarios 11 Table 3. (Continued) 12 Table 4. Sectoral GDP growth results for the growth scenarios 14 Table 5. Factor market results for growth scenarios 15 Table 6. Household results for growth scenarios 16 Table A1. CGE model sets, variables and parameters 22 Table A2. CGE model equations 23 Table A2 continued. CGE model equations 24 Table A2 continued. CGE model equations 25 Table A3. Regional and international trade elasticities in the model 26 iv

5 ACKNOWLEDGEMENTS We are grateful to Maximo Torero (IFPRI) for comments, and to participants of workshops held at Grupo de Análisis para el Desarrollo and the Inter-American Development Bank in Lima, Peru. We are also thank the Inter-American Development Bank for funding this research. v

6 ABSTRACT Despite the economic transformation of Peru s coastal economy, the country s inland region remains poor and underdeveloped. We herein examine the economic linkages between the two regions using a multi-regional computable general equilibrium model based on a regionalized social accounting matrix. The model results show that coastal growth undermines the inland economy by increasing import competition and internal migration. Peru, therefore, cannot rely solely on rapid national growth to generate broad-based poverty reduction. When we simulate policies aimed at curbing divergence, we find that reducing interregional transaction costs stimulates national economic growth, but widens divergence by shifting inland production towards agriculture and concentrating investment in coastal manufacturing. In contrast, conditional cash transfers reduce regional and rural-urban inequality, but do not stimulate national growth. Finally, investing in inland productivity (through extension services and improved rural roads) reduces regional divergence, but the resulting market constraints worsen rural-urban inequality. These findings suggest that isolated interventions may worsen inequality, and that complementarities exist between supply-side investments and policies aimed at stimulating demand and improving access to national markets. Keywords: regional development; public investments; economic growth; Peru vi

7 1. INTRODUCTION The persistence of rural poverty in Peru s inland Sierra and Selva regions is one of the country s most pressing socioeconomic problems. Peru made some progress in reducing national poverty during the 1990s, but this was later reversed by the economic slowdown during (World Bank, 2005). The Peruvian economy has boomed since then, with national gross domestic product (GDP) growing at over 5 percent per year and poverty declining significantly. This recent expansion has been driven by particularly strong growth in mining and manufacturing (INEI, 2008). However, employment creation is still dominated by low-paying informal services (World Bank, 2005) and poverty remains high, with twofifths of the population living below the poverty line in 2007 (INEI, 2009). Moreover, the gains from this rapid economic growth have been largely concentrated in urban areas and along the coast, with poverty, particularly extreme poverty, declining only marginally in the rural inland regions. Inland poverty would be a less serious problem for the country as a whole if the population of these regions were small. However, the Sierra and Selva regions contain half of Peru s total population and 90 percent of its extremely poor population (INEI, 2006). The problem would also be less serious if these regions were catching up with the coastal economy. However, there is little evidence to suggest that regional convergence is taking place. Currently Peru s economy is expanding at over 7 percent per year, but the problem of the country s backward regions persists. Indeed, the very success of the coastal agroindustrial sectors is widening regional inequality, increasing divergence and exacerbating social and political pressures in and between the regions. What is striking about this recent boom period is the difference in poverty trajectory between the urban and rural sectors, in particular the gap that is emerging between Lima on the ond hand and the Coast and Sierra-Selva regions on the other. 1 For instance, the overall poverty rate in the rural sector of Peru increased during the boom, largely because the poverty reductions in the Coast and Selva regions were offset by poverty increases in the more populous rural sector of the Sierra region. Indeed, overall poverty rates fell by 35.6 percent on the coast between 2004 and 2007, but by only 9.4 percent in the Sierra and Selva regions. Moreover, poverty rates in the inland region remained virtually unchanged during the first two years of the economic boom (i.e., ), while they declined dramatically along the coast. The contrast between the regions is even more evident for extreme poverty, which rose in the Sierra region during In short, poverty in the coastal region has proven to be highly sensitive to the rate of economic growth, whereas inland poverty has not (World Bank, 2005). Thus, Peru has a structural lagging regions problem common to many fast-growing developing countries. It has a geographic area containing a significant share of the population, whose development has systematically lagged behind that of the rest of the country. The underlying mechanics that have generated growth at the national level in the past have not improved incomes for the poor in Peru s inland regions. This raises two important developmental issues. First, we can conclude that the poor in the Sierra and Selva regions are either not well linked to the more advanced coastal economy, or they are linked in a way that reinforces regional divergence. Second, as a result of these weak or adverse regional linkages, it is unlikely that Peru will be able to solve its rural poverty problem by focusing on accelerating its national growth rate. Peru is a clear example of a case in which growth does not trickle down to the poor. For this reason, policy interventions or changes in the country s growth strategy will be required to improve the linkages between the inland and coastal economies and allow the poor to benefit more from the growth process. In this paper, we address both of these development issues. In Section 2, we measure the linkages between the inland and coastal economies using regional social accounting matrices (SAM) and evidence from the literature. Drawing on this information, we construct a dynamic multi-region computable general equilibrium (CGE) model, which is described in Section 3. In Section 4, we use the model to consider how the recent boom in Peru s coastal economy affects the inland economy. We find that, while coastal 1 For an analysis of the regional poverty statistics between 1997 and 2004, see Escobal and Valdivia (2004); for the period, see INEI,

8 growth increases household incomes, it has adverse effects on regional and rural/urban divergence. In Section 5 we examine three possible policy responses to this divergence. We first consider the effects of investing in interregional road infrastructure, thereby reducing the remoteness of Peru s lagging region. Secondly, we consider the impact of increasing social transfers to inland households and raising productivity in the region. Finally, we consider investments aimed at improving agricultural productivity in the inland and coastal regions. These three scenarios reflect alternative uses of public resources to address the concentration of growth and poverty reduction within the coastal region, and are therefore relevant to the allocation decisions underlying the current system of regional public transfers. The final section draws conclusions for the lagging region problem in Peru and elsewhere. 2

9 2. REGIONAL LINKAGES IN PERU Uneven development is characteristic of rapid growth episodes in most countries. However, lagging regions may become a national development problem when regional inequalities persist over a long period of time despite the occurrence of structural transformation elsewhere in the economy. Regions may initially fall behind for a number of reasons, including climatic handicaps, ethnic differences, regionally discriminatory policies, and civil wars. Regional divergence may continue even after the initial differences become less binding, due to the agglomeration of economic activity and the resulting pathdependency of the development process. In Peru, mountains form a natural division between the narrow coastal region and the inland Sierra and Selva regions. While the coast contains half of the country s population and two-thirds of its urban population, the inland region contains most of the rural population, and a large proportion of the poor and extremely poor. Given this sharp regional division, it might seem appropriate to treat the lagging and leading regions as isolated countries, thus implicitly assuming that there are no linkages between them. In this section, we consider the appropriateness of this assumption. Drawing on a multi-regional SAM constructed by Morley et al. (2008), we examine four kinds of linkages: production and trade flows; labor markets and internal migration; capital markets and investment; and public spending and social transfers. Production and Trade Linkages The coastal and inland regions have very different economic structures. The coastal economy, which accounts for 70 percent of national GDP, is dominated by urban-based manufacturing and private services (see Table 1). Agriculture is less important and consists mainly of larger-scale commercial enterprises. In contrast, the inland regions are heavily dependent on agriculture, most of which is undertaken by smallholder farmers. Although mining and the refinement of non-ferrous metals are major sectors in the inland region, most heavy manufacturing takes place on the coast. Inland manufacturing, on the other hand, consists mainly of light industries, such as food processing, textiles and wood products. Despite its smaller economy, however, the inland region accounts for almost half of Peru s foreign earnings. This can be seen in the macro SAM for the two regions (see Table 2). 2 In the table, foreign exports are payments from the rest of the world for locally produced commodities, while exports appear in the rest of world column and the commodities row (i.e., coastal and inland foreign exports equaled 18.2 and 14.6 billion soles, respectively, in 2002). The inland region s large contribution to exports is primarily through mining and metals, although coffee, fish and wood products are also important inland exports. Thus, while the inland region runs a foreign trade surplus equal to 11 percent of its GDP, the coastal region runs an equally large trade deficit. This is the first trade-related linkage between the two regions, and comes through their contributions to a common current account balance. If the two regions were separate countries, then the inland region could offset its competitive disadvantage through adjustments to its exchange rate policy. By devaluing its exchange rate, the inland region could improve the competitiveness of its foreign exports. However, because the regions are part of a single country with a common trade policy, changes in the terms-of-trade or export performance of one region will influence the other region by inducing changes in Peru s exchange rate. 2 A SAM is a consistent framework that captures all of the economic flows in an economy. The SAM is a square matrix, in which rows represent receipts and columns are payments. The Macro SAM in Table 2 is an aggregate version of the detailed SAM described in Morley et al. (2008). 3

10 Table 1. Summary statistics of coastal and inland regions Year National Coastal Inland Both Rural Urban Both Rural Urban Population (1000s) ,420 13,109 1,259 11,850 13,311 7,224 6,086 Poverty rate (%) Extreme poverty rate (%) Expenditure per capita (S/p.a.) ,414 7,627 3,212 8,096 3,234 1,592 5,182 GDP per capita (S/p.a.) ,798 9, ,096 GDP at factor cost (S/mil) , , , Agriculture 13,962 4, , Mining 8,190 2, , Manufacturing 28,246 20, , Other industry 15,404 8, , Services 113,789 89, , Dependency ratio (people/worker) Employment (1000s) ,879 5, ,905 4,532 2,126 2,406 Skilled 1, Semi-skilled 3,764 2, ,455 1, Unskilled 4,967 2, ,692 2,917 1,872 1,045 Wage rates (S/per worker p.a.) ,174 10, , Skilled 15,918 18, , Semi-skilled 10,981 11, , Unskilled 4,258 6, , Regional trade surplus , , Foreign trade surplus ,692-11, , Fiscal surplus ,650 2, , Source: Population is from 2004 ENAHO (INEI, 2006); poverty rates are from INEI (2009); data on expenditures, GDP and trade and fiscal surpluses are from the 2002 Peru SAM (Morley et al., 2008); employment, wages and dependency ratios are the authors calculations using the 2002 Peru SAM and 2004 ENAHO. Note: Labor skills are based on occupational categories (skilled includes professional and managers; semi-skilled includes formal sales workers and machinery operators; unskilled includes agricultural and domestic workers and informal retailers). We assume that rural households have (at most) semi-skilled labor. 4

11 Table 2. Macro social accounting matrix (SAM) for Peru s coastal and inland regions (2002, millions of Soles) Activities Commodities Factors Households Government Taxes Savings Regional linkages Rest of World Total Activities 96,053 46, ,069 54, , ,699 Commo dities 221, ,699 15,104 4,176 9,295 23,197 28,270 6, , ,377 Factors 120,124 50,024 4,465 1,037 1,634 2, ,223 54,523 99,984 43,045 1, ,720 8, ,413 52,181 12,932 7,302 6,288 2,157 2,057-3,707 21,278 5,751 Taxes 21,278 5,751 21,278 5,751 Households Government Investment 23,409 13,950 23,409 13,950 Source: Morley et al. (2008) Note: The top number in each cell represents the coastal region and the bottom number represents the inland region Regional linkages 23,197 9,295 1, ,056 15,056 9,295 24,351 Rest of World 18,216 14,610 11,688-5,996 29,904 8,614 Total 221, , , , ,223 54, ,413 52,181 21,278 5,751 21,278 5,751 23,409 13,950 9,295 24,351 29,904 8,614 5

12 The SAM also captures interregional trade between the two regions. These are payments by one region for the other region s commodities (i.e., the regional linkages row and column in Table 2, representing regional exports and regional imports, respectively). The macro SAM shows that interregional trade is as important as international trade for the two regions: first, the inland region imports most of its manufactured goods from the coast; and second, international imports must travel through the coastal region to reach the inland region. As such, the inland region relies heavily on imported trade services, which are embodied in its imported manufactures. This results in high transaction costs between the coastal and inland regions. Overall, the inland region s regional trade deficit is equal to 27 percent of its GDP. This is partially offset by agriculture-related exports to the coast (e.g., horticultural products, livestock and wood products). These exports supply around half of the coastal consumption of these goods. Since the regional trade balance is far larger than most countries external balances, it is clear that there are substantial interregional trade linkages in Peru despite the country s difficult terrain. Production changes in the coastal region will therefore have significant implications for producers and consumers in the inland region. If the inland region were a separate country, it could adjust its trade policies. For example, the inland region could protect local producers by imposing tariffs on coastal imports. However, there are no sub-national trade policies in place at present, and domestic producers face regional competition (albeit constrained by high interregional transaction costs). Capital Markets and Investment The second linkage between the regions is in the capital market. Both regions generate savings and make investments. The SAM does not have a regionalized financial sector, which is equivalent to assuming that all savings go into a national pool to be allocated to investment as determined by relative profit rates. In essence, all sectors in both regions compete for the national supply of savings, which means that if profit rates are higher on the coast, inland savings should be drawn out of the region, thus exacerbating tendencies toward divergence. Conversely, if returns are higher in the inland region, then investment could be drawn away from the coast, thereby helping to reduce divergence between the regions. The investment column in the macro SAM distributes total investment across the two regions, while the savings row identifies how investment is financed. Under the ex post equilibrium condition, total investment equals total saving in each region. As mentioned above, the coastal region runs a large trade surplus with the inland region and a smaller trade deficit with the rest of the world. Coastal households also generate a larger share of domestic private savings. To some extent, this explains the larger allocation of investment to the coastal region. However, investment is not allocated to regions according to savings contributions. The macro SAM shows that the coastal region financed a quarter of inland investment during If the two regions were separate countries, they could use capital controls to prevent large outflows to other regions. However, as regions within a country and given Peru s established financial systems, there are few policies and barriers to prevent the inland region s profits and savings from being reinvested in the coastal region where returns may be higher. Conversely, national capital markets may provide the financing needed to allow investment to rise beyond regional savings. Finally, it is worth noting that investment responds to sectoral or firm-level profit differentials. For this reason, while the inland region is dominated by land-intensive agriculture, the region may still attract a larger share of investment due to higher returns to more capital-intensive mining and manufacturing. It is therefore more difficult to predict sectoral capital movements than it is to predict regional labor migrations. Public Sector and Social Transfers Government is a potential channel for reducing regional inequality, through either targeted expenditures or social transfers. Governments raise revenues through taxes and use these revenues to finance public investments, social transfers, and recurrent expenditures. Due to its low-income population and small formal sector, the inland region ran a large fiscal deficit during 2002 (equal to 6.8 percent of its GDP). Conversely, the coastal region raised more tax revenues than were spent in the region, and therefore financed a large share of the inland region s fiscal deficit. These transferred tax revenues contributed to the provision of social services (e.g., education and health) in the inland region; in 2002, they were equal 6

13 to the social transfers paid directly to inland households. However, despite the interregional subsidization of public services, the coastal region still attracted more than 70 percent of public sector spending in that year, a large proportion of which was directed towards the central government. Only 8 percent of national tax revenues were redistributed from the coast to the inland region. Thus, compared to interregional trade, the reallocation of public funds is a less significant source of regional linkages. However, in the absence of other possible policy interventions, the public sector is currently the primary mechanism for stimulating regional growth and convergence. Labor Markets and Internal Migration The fourth linkage between the two regional economies is the migration of labor from inland regions to the coast. Employment in the inland region primarily consists of lower-skilled jobs, reflecting the importance of agriculture. Most better-paying semi-skilled jobs are in manufacturing, while skilled labor is almost exclusively employed in the public sector. In contrast, the coastal economy generates far more employment opportunities for semi-skilled workers. There are also large between-region differences in average wages. This is especially true for lower-skilled workers, with inland workers earning less than half what workers in similar occupations earn in the coastal region. Semi-skilled wage differentials are narrower, due in large part to the existence of higher-paying mining and public sector jobs in the inland region. Finally, dependency ratios are also higher in the inland region, suggesting that not only do inland workers tend to have lower skills and wages, each worker s income also typically supports more dependents than is the case in coastal households. This accounts for some of the differences in per capita expenditure between the inland and coastal regions. Evidence suggests that workers respond to relative wage differentials by moving over time from low-wage occupations in the inland region to higher-wage labor markets on the coast (Laszlo and Santor, 2004). A key empirical question is what fraction of the low-wage region s labor force moves per time period. If the fraction is large, then the regional disparities in earnings are likely to be transitory. However, if the fraction is small, then large income disparities may persist. Furthermore, it may seem obvious that migration is a dynamic equilibrating mechanism, since the movement of workers will tend to reduce regional wage differentials. However, since the rate of return to capital is a positive function of the supply of labor, out-migration from the inland region will also tend to increase the rate of return to capital on the coast. Since the regional allocation of investment is a function of the relative rates of return, both investment and the capital stock will tend to move toward the area that receives migrants and away from the sending region. This exacerbates differences in regional growth rates and further highlights the importance of regional linkages. 7

14 3. MODELING THE LINKAGES BETWEEN THE COASTAL AND INLAND REGIONS The discussion in the previous section suggests that it would be incorrect to treat Peru s lagging region as an isolated economy or a separate country. There are numerous linkages between the coastal and inland regions, some of which exist because these regions are part of the same country. Here, we construct a multi-regional recursive-dynamic CGE model designed to capture the linkages described in the previous section. This section describes the structure and behavior of the model, while a detailed mathematical specification is provided in the appendix. A Two-Region CGE Model of Peru The coastal region contains most of Peru s manufacturing and private services, while the inland region depends more on agriculture and mining. To capture this heterogeneity, the model contains detailed information on the demand and supply of 37 economic sectors/commodities in each of the two regions. 3 Producers employ land, labor and capital under the assumption of constant returns-to-scale and profit maximization. For this, we use a nested production system, with a constant elasticity of substitution (CES) function that determines factor demand and a Leontief function that combines value-added and intermediates. The model separates skilled, semi-skilled and unskilled workers, which are used with differing intensity in each sector. We assume that workers within each region migrate between sectors according to labor demand, and that Peru s total labor supply grows at a fixed rate (i.e., national unemployment rates remain constant). Agricultural producers use unskilled labor, region-specific land, and (to a lesser extent) capital. Nonagricultural producers also use labor and capital, although the mining sector has its own capital stock. All existing capital is immobile across sectors, earning sector-specific profits. The detailed specification of production and factor markets in the model allows it to capture the unique structure of these two regions economies. The first regional link is trade. The model explicitly allows for interregional and international trade. Import competition and export opportunities are captured by allowing producers and consumers in each region to shift between regional and foreign markets depending on the relative prices of imports, exports and local goods. More specifically, the decision of producers in each region to supply local, regional or foreign markets is governed by a constant elasticity of transformation (CET) function, while substitution possibilities between local and imported goods are captured with a CES Armington function. Although this specification allows for two-way trade between the two domestic regions, it is only possible to estimate net flows. This means that if a region is initially a net importer of a commodity, then it cannot later become a net exporter. In most cases, this is not an unreasonable assumption. For example, natural conditions make it unlikely that the coastal region could become a net exporter of agricultural and mining goods to the inland region. Increased coastal production can, however, expand its exports to international markets and/or substitute for imports from the inland region. Finally, the model also captures interregional transaction costs, which are imposed on all goods entering or leaving the inland region. We assume that this additional cost of regional trade generates demand for the exporting region s trade sector. Therefore, by explicitly allowing for interregional trade and transaction costs, the model captures the commodity market linkage between the two regions. Household income and expenditure patterns vary considerably across regions and between rural and urban areas. These differences are important, since the incomes earned by workers in different sectors will benefit households differently according to their location and factor endowments. To capture these differences, the model separates rural and urban households in each region. These representative households receive factor incomes and per capita transfers from the national government. Rural households receive most of their incomes from land and lower-skilled workers, while urban households receive a greater share from non-mining capital and higher-skilled workers. All households save some of their incomes (based on fixed marginal propensities to save), but only urban households pay taxes (based 3 The model focuses on the agricultural and manufacturing sectors, which have 15 and 12 sub-sectors, respectively. However, the model captures all sectors, including mining, construction, utilities, and various private and public services. For more details on the database, see Morley et al. (2008). 8

15 on fixed tax rates). Each household uses its remaining income to consume commodities under a linear expenditure system (LES) of demand. The second regional link is internal labor migration. The model allows workers in each region to migrate to the other region if relative wages are higher. However, the labor markets in the model are imperfect, and regional wages may not equalize over the medium-term. Using a linear migration function, we assume that if the coastal wage differential rises by 1 percent relative to the base period differential, then there is a 0.2 percent net out-migration of inland workers to the coast. As mentioned in the previous section, large wage differentials already exist between the coastal and inland regions. To some extent, this reflects differences in regional living costs and the socioeconomic costs of migration. We therefore assume an equilibrium-compensating wage, such that long-run factor mobility tends to the initial wage differential rather than equalization. The model also captures the influence of migration on population growth. Each migrating worker co-migrates with a third of the dependents from their representative household. This assumption is consistent with the observation that younger workers with smaller families more often migrate, leaving an aging non-working population in the inland region. Since there is no demographic model embedded in the CGE model, we assume that the national population grows in line with the labor force (i.e., the national dependency ratios remain constant). Finally, the allocation of new migrants between rural and urban areas occurs in proportion to the working population of the home and destination regions. This specification of migration and population growth allows the model to capture labor market linkages between the two regions and the demographic effects of this linkage. The third regional link is the public sector. Being connected to a larger economy allows the inland region to run a larger recurrent fiscal deficit than would be possible if it were a separate country. In the base period, the inland region receives a larger share of government spending than its share of tax revenues. The model captures this by pooling all tax collections at the national level, including regionspecific sales and income taxes and import tariffs. Public sector borrowing, which is a fixed share of public revenues, is also pooled. The national government first uses these revenues to fund per capitabased household transfers in each region. The remaining funds are then used for recurrent consumption expenditures, which are allocated in fixed proportions across regions. 4 Therefore, regional tax collections vary with each region s GDP, but regional public spending is determined by population growth and past expenditure patterns. The latter include the current system of regional transfers to local municipalities (e.g., the canon). The fourth and final regional link is savings and capital investment. As discussed above, it is possible for a region to receive a larger share of investment than its contribution to savings if producers in that region are able to earn a higher return on capital. The model captures this as follows. First, all public and private savings and foreign borrowing are pooled at the national-level, thus determining the total amount of investment spending in the economy. Second, investment spending is allocated across sectors and regions according to profit-rate differentials. Sectors and regions with above-average returns receive larger shares of new capital, which is needed to augment depreciating stocks. Capital accumulation is therefore endogenous and investment spending is determined at the sectoral level. Under this specification, inland savings may be invested in the coastal region if profit rates are higher there. Conversely, the inland region can attract investment beyond the level that can be financed by the region s own savings. Calibrating the Model The main database used to calibrate the model is a 2002 multi-regional SAM for Peru (Morley et al., 2008). The SAM captures the structure of the country after the economic crisis and before the recent growth acceleration. In the SAM, data on regional production come from agricultural surveys and national accounts. Information on labor markets and household income and expenditure patterns are taken from the 2004 Encuesta Nacional de Hogares (a nationally-representative household survey, see INEI, 2006). Interregional trade flows are calculated as a residual, after reconciliation of regional production 4 Notably, while the government s recurrent consumption spending is exogenously allocated across regions based on past trends, public investments are endogenously determined alongside private investments (i.e., based on regional profit differentials). 9

16 and demand data. The model also contains a number of elasticities. Trade elasticities are taken from Dimaranan et al. (2006), 5 while commodity-specific income elasticities are econometrically estimated for rural and urban households using data from the 2004 household survey. 5 The Global Trade Analysis Project s (GTAP) model estimates lower elasticities for manufactures and traded services, and higher elasticities for agricultural goods, which are more disaggregated or homogenous in both the GTAP and the current model. 10

17 4. SOURCES OF GROWTH AND DIVERGENCE Recent growth in Peru has reduced poverty, but mainly along the coast. This implies that growth in the coastal region may not stimulate growth throughout the economy. The previous sections highlight the linkages between the two regions. In this section, we use the CGE model to examine the implications of these linkages. Scenario 1: A Balanced Growth Baseline Scenario We initially calibrate the model to produce a baseline scenario in which both regions grow equally fast. We run the model forward over 10 periods while updating exogenous parameters. For convenience, we refer to these time periods as years. We assume that the total supply of land and skilled, semi-skilled and unskilled labor grows at 2 percent per year. We also calibrate investment prices such that capital stocks grow at roughly 2 percent per year (after applying a 5 percent depreciation rate). Finally, we impose 1 percent annual total factor productivity (TFP) growth in all sectors in both regions. Together, 2 percent factor growth and 1 percent TFP growth yield a national GDP growth rate of around 3 percent (see Table 3). GDP growth deviates only slightly between the two regions. This is due to differences in their economic structures and elasticities. For example, the income elasticities for the kinds of goods produced in the inland region are typically lower than those for most coastal goods. Since growth is fairly balanced across the regions, there is no significant shift in wage differentials and few incentives for labor to migrate between regions. Accordingly, employment grows at about 2 percent in both regions, which also means that the population shares remain unchanged. Thus, under the baseline scenario, we do not see significant regional or divergence. Table 3. Macroeconomic results for growth scenarios Initial values Baseline scenario (1) Coastal boom (2) Trans. costs (3) Social transfer (4) Inland prod. (5) Final period values Consumer prices (index) Coastal Inland Real exchange rate (index) Average growth rate (%) Total GDP (2002 soles) 179, Coastal 125, Inland 54, Investment demand 37, Coastal 23, Inland 13, International exports 32, Coastal 18, Inland 14,

18 Table 3. (Continued) Initial values Baseline scenario (1) Coastal boom (2) Trans. costs (3) Social transfer (4) Inland prod. (5) International imports 34, Coastal 28, Inland 6, Regional exports Coastal 23, Inland 9, Percentage point contribution to GDP growth rate Coastal GDP (% ) Labor employment Capital stocks Land area TFP Inland GDP (% ) Labor employment Capital stocks Land area TFP Source: Peru CGE model results. Note: The exchange rate index is local currency units per foreign currency unit (increase is an appreciation). The consumer price index is relative to the inland region s domestic price index (i.e., the model s numeraire). Scenario 2: Faster Nonagricultural Growth along the Coast Manufacturing and private services tend to dominate growth in Peru, accounting for 36.5 and 48.2 percent of GDP growth, respectively, during (INEI, 2008). Most of this growth has taken place within urban centers along the coast (Giugale et al., 2007). Thus, in the second scenario we simulate an expansion of the coastal economy by increasing nonagricultural TFP growth from 1 to 2.5 percent per year (excluding the mining sector). This productivity increases the coastal GDP growth rate from 3.1 percent in the baseline scenario to 4.2 percent per year, which is larger than the increase in TFP (see Table 3). Moreover, GDP growth in the inland region falls from 3 to 2.7 percent per year. To understand why faster coastal growth comes at the expense of inland growth, we consider the various linkages discussed in the previous sections. Higher productivity in the coastal region stimulates production, particularly in manufacturing, which sees a growth rate increase of an additional 2.2 percentage points (see Table 3). Expanding production causes coastal prices to fall relative to inland prices. As a result, exports from the coast to the inland region increase by an additional 0.8 percentage points per year, mainly through manufacturing (see Table 4). Increased competition undermines growth in the inland region s heavier manufacturing sectors and causes overall manufacturing to contract relative to the Baseline scenario. However, faster growth and higher incomes in the coastal region increase the demand for commodities produced in the inland region. Unlike the coastal region, though, the increase in the inland regional exports comes at the expense of its foreign exports. Inland producers, who are not more productive in this scenario, shift their production away from foreign markets towards domestically oriented agricultural sectors (e.g. livestock 12

19 and forestry) and light manufacturing (e.g. textiles and clothing). Ultimately, faster growth in the coastal region encourages interregional trade, but undermines inland manufacturing and reduces export opportunities. It also increases the inland region s dependency on agricultural production. 13

20 Table 4. Sectoral GDP growth results for the growth scenarios Initial GDP values Baseline average Percentage point deviation from baseline growth rate (2002 soles) growth (1) Coastal boom (2) Trans. costs (3) Social transfer (4) Inland prod. (5) Coast Inland Coast Inland Coast Inland Coast Inland Coast Inland Coast Inland Total GDP 125,069 54, Agriculture 4,407 9, Cereals 520 1, Roots 159 1, Horticulture 1,604 3, Coffee Other export 855 1, Livestock 483 1, Forest Fish Mining 2,174 6, Manufacturing 20,056 8, Food 7,253 2, Textiles 2,689 1, Wood Chemicals 4, Non-metals 1, Metals 1,389 1, Machinery Other 1, Construction 7,313 4, Utilities 1,504 2, Trade/transport 40,988 9, Regional trade 634 2, Private services 38,649 6, Public services 9,343 5, Source: Peru CGE model results. 14

21 Rising productivity increases the returns to capital in the coastal region (see Table 5). This allows coastal sectors to attract a greater share of new capital investment, thereby displacing investment in the inland region (see Table 3). Faster accumulation of capital in the coastal region also increases the demand for labor and places greater upward pressure on coastal wages. This widens the regional wage differentials and encourages workers to migrate to the coast (see Table 5). Out-migration is especially pronounced for semi-skilled workers, who are used more intensively in the coastal region s nonagricultural sectors. This contributes to a decline in inland manufacturing and private services. In contrast, there are relatively fewer job opportunities for lower-skilled workers in the coastal region. Although this dampens some of the out-migration of these workers, it further encourages a shift towards inland agriculture and lower-skilled manufacturing. Thus, labor migration and slower capital accumulation contribute to the inland economy s decline by reducing the productive capacity of the region. Table 5. Factor market results for growth scenarios Initial values Baseline scenario (1) Average annual growth rate Trans. costs (3) Coastal boom (2) Social transfer (4) Inland prod. (5) Skilled labor wages (2002 soles) Coast 18, Inland 11, Semi-skilled labor wages (2002 soles) Coast 11, Inland 9, Unskilled labor wages (2002 soles) Coast 6, Inland 2, Capital returns - Coast Inland Land rental rate - Coast Inland Skilled labor employment (1000s) Coast Inland Semi-skilled labor employment (1000s) Coast 2, Inland 1, Unskilled labor employment (1000s) Coast 2, Inland 2, Capital stock - Coast Inland Source: Peru CGE model results. 15

22 Although GDP falls in the inland region, there is higher overall growth at the national level. This raises national tax revenues and permits an expansion of the public sector in both the coastal and inland regions (see Table 3). Therefore, with falling or stagnant growth in most inland sectors, the public sector becomes an important source of inland growth and employment. This offsets some of the migration of higher-skilled workers from the inland region. Moreover, while the out-migration of workers causes inland dependency ratios to rise (see Table 6), it does not affect the provision of per capita-based social transfers, which are mainly financed by coastal taxes. Thus, with a booming coastal region, the public sector becomes an increasingly important part of the inland region s economy. Table 6. Household results for growth scenarios Initial values Baseline scenario (1) Coastal boom (2) Average growth rate (%) Trans. costs (3) Social transfer (4) Inland prod. (5) Population (1000s) 26, Coastal 13, Inland 13, Dependency ratio (people) Coastal Inland Per capita equivalent variation (2002 soles) Coastal rural 3, Coastal urban 8, Inland rural 1, Inland urban 5, Source: Peru CGE model results Note: Dependency ratio is the number of non-working people per worker. Equivalent variation is a welfare measure (i.e., additional income required to increase household utility to its final year value based on initial prices). Initial values are annual per capita expenditures. The above results suggest that regional trade and factor market linkages are not weak. Rather, they act against growth in the inland region, with the public sector only partially offsetting these adverse linkages. This implies that there is greater economic divergence as a result of the coastal and inland regions being part of the same country. However, while GDP declines in the inland region, its factor returns increase in real terms due to decreases in local prices and the falling cost of goods imported from the coast (see Table 5). This allows real incomes to rise, especially for rural households, which also benefit from increases in the food crops, livestock and textile sectors. Accordingly, most of the income growth among rural inland households comes from higher returns to land and unskilled labor. However, coastal households see greater income increases, driven by increased employment opportunities and falling consumer prices. In summary, all households in Peru benefit from faster growth in the coastal economy. However, there is a widening income gap between coastal and inland households, and greater regional divergence in economic growth. Furthermore, there is an increased specialization of the inland economy in agriculture and public services. This suggests that, despite positive public sector linkages, regional trade and migration linkages reinforce slower growth in the Sierra and Selva regions compared to the coastal region. This is consistent with the observed divergence in regional incomes and the persistence of inland poverty despite the accelerated economywide growth seen in recent years. 16